Goal‐driven Autonomy for Responding to Unexpected Events in Strategy Simulations

Abstract

To operate autonomously in complex environments, an agent must monitor its environment and determine how to respond to new situations. To be considered intelligent, an agent should select actions in pursuit of its goals, and adapt accordingly when its goals need revision. However, most agents assume that their goals are given to them; they cannot recognize when their goals should change. Thus, they have difficulty coping with the complex environments of strategy simulations that are continuous, partially observable, dynamic, and open with respect to new objects. To increase intelligent agent autonomy, we are investigating a conceptual model for goal reasoning called Goal‐Driven Autonomy (GDA), which allows agents to generate and reason about their goals in response to environment changes. Our hypothesis is that GDA enables an agent to respond more effectively to unexpected events in complex environments. We instantiate the GDA model in ARTUE (Autonomous Response to Unexpected Events), a domain‐independent autonomous agent. We evaluate ARTUE on scenarios from two complex strategy simulations, and report on its comparative benefits and limitations. By employing goal reasoning, ARTUE outperforms an off‐line planner and a discrepancy‐based replanner on scenarios requiring reasoning about unobserved objects and facts and on scenarios presenting opportunities outside the scope of its current mission.

Robot task planning and explanation in open and uncertain worlds

“…In experiment 4, all the suitable runs that failed in experiments 1-3, plus seven additional runs with forced failures, were given to the planner to explain the failures using diagnostic knowledge. This experiment shows the ability of the system to use a mix of default assumptions and instance assumptions to create 23 See http://openmind.hri-us.com/. This was obtained from humans freely listing object types they associated with given room types.…”

“…All these frameworks deal with identifying a set of propositions or beliefs that either lead to inconsistencies or permit the robot to deduce an observation, while we have a set of operators that allow us to transform states. The idea of detecting discrepancies between expected and observed state also comes up in the goal-driven autonomy framework of Klenk et al [23], where discrepancies trigger the generation of explanations, which in turn lead to the creation of new goals. There is also a relationship to work in explanation-based learning [30].…”

“…of possible object-room relations was extracted from the Open Mind Indoor Common Sense (OMICS) database. 23 . Probabilities for relations were estimated from image search-engine queries on the Web [13].…”

Case-Based Learning of Applicability Conditions for Stochastic Explanations

“…More recently, Klenk et al (2012) presented ARTUE, which uses a direct application of explanations in a strategy simulation. ARTUE is based on a modified version of the SHOP2 planner, and it tries to explain discrepancies similarly to DiscoverHistory.…”

Multi-layered vehicle control via payload autonomy

“…This is especially true as development moves into higher levels of cognition and reasoning such as goal-based autonomy [16]; developers should be able to use as many existing capabilities as possible to minimize the additional work required.…”